The Autonomous Quantum Chip Fab of 2035

下午 4:50 - 下午 5:10

Quantum chip production is transitioning from artisanal, lab-scale proof-of-concepts toward engineering-driven commercial manufacturing, mirroring the semiconductor industry's move toward autonomous, "lights-out" foundries governed by model-based, closed-loop control with minimal human intervention.

Reconciling these trajectories is non-trivial: quantum chip fabrication inherits only some of the deterministic process–performance correlations classical fabs rely on. True performance metrics such as coherence times and gate fidelities are extractable only under cryogenic conditions, at high cost and late in the flow, leaving processes severely constrained in both observability and controllability.

The first step is a consistent, traceable data infrastructure that statistically links fabrication metadata, test results, and environmental conditions, with causality and measurement uncertainty captured explicitly.

A first version of this data management infrastructure which establishes monitoring and control for an entire end-of-line testing system is deployed in the existing OrangeQS MAX system.

We seek partners of multiple qubit modalities within the partnership program to expand the MAX line and the data management infrastructure. This allows both a combination of AI and physics-based models to act as "virtual sensors" for closed-loop control across different timescales and various types of qubits whenever a measurement is available.

Featured Speakers

Kelvin Loh

Lead Product Manager & Co-founder, OrangeQS

Kelvin co-founded Orange Quantum Systems in 2020 as a QuTech spin-off. An applied mathematician by training with a focus on high-performance computing, he led the company's software development from 2020 to 2022 as maintainer of the open-source Quantify framework and the Superconducting Qubit Tools. From 2022 to 2024, as interim Commercial Director, he drove the first sales across the MAX and FLEX product lines. Since 2024 he is the Lead Product Manager for the MAX system, having deployed the OrangeQS MAX system to industrial customers and leading the partnership program to develop the next generation of high-throughput test systems.

Before OrangeQS, Kelvin worked at TNO, where his work bridged machine learning and physics-based modeling. He developed the first data assimilation approach coupling an ensemble Kalman filter with a physics-based LSTM network, applied to real gas fields; and performed charge stability diagram simulations using customized coupled TCAD and Schrödinger-Poisson FEM solvers for real double quantum dot devices using actual SEM geometry, demonstrating a feasible closed-loop feedback development cycle for spin qubit devices.